delta/qt5/qtwebkit.git/Source/JavaScriptCore, branch old/5.2

[mips] Wrong register usage in LLInt.

2014-02-10T15:27:56+00:00

Fix register usage and add PIC header to all LLInt operations.
This patch is taken from https://bugs.webkit.org/show_bug.cgi?id=125168.

It fixes the crash of fast/js/exception-propagate-from-dfg-to-llint test
on mips architecture.

Change-Id: I98ad3b5766451cab48a76f7e028b210f9ebe99ed
Reviewed-by: Allan Sandfeld Jensen

Do not completely erase LIBS variable

2013-12-17T16:08:07+00:00

Using an empty LIBS variable prevents usage of custom library paths
and libraries via configure -L  -l .
This is needed for linking with an alternative stdlib implementation.

Change-Id: Ie1009bfd435436bf584b2963066535ee90f4d5d0
Reviewed-by: Pierre Rossi 
Reviewed-by: Allan Sandfeld Jensen

Merge remote-tracking branch 'origin/release' into stable

2013-11-26T09:54:43+00:00

Change-Id: I9088052f76de19e584cd16fd039502de6b22bec2

[sh4] Fix revertJumpReplacementToBranchPtrWithPatch in MacroAssembler.

2013-11-20T14:02:31+00:00

https://bugs.webkit.org/show_bug.cgi?id=124468

Patch by Julien Brianceau  on 2013-11-17
Reviewed by Michael Saboff.

Current implementation of revertJumpReplacementToBranchPtrWithPatch is wrong in
the sh4 MacroAssembler part, leading to random instabilities. This patch fixes it
and also renames the bad-named revertJumpToMove to revertJumpReplacementToBranchPtrWithPatch
in the SH4Assembler.

* assembler/MacroAssemblerSH4.h:
(JSC::MacroAssemblerSH4::revertJumpReplacementToBranchPtrWithPatch):
* assembler/SH4Assembler.h:
(JSC::SH4Assembler::replaceWithJump):
(JSC::SH4Assembler::revertJumpReplacementToBranchPtrWithPatch):

Change-Id: Ifbc6851f506a1303c85cff3a47c23e932fad033c
git-svn-id: http://svn.webkit.org/repository/webkit/trunk@159400 268f45cc-cd09-0410-ab3c-d52691b4dbfc
Reviewed-by: Allan Sandfeld Jensen

[sh4] Protect repatchCompact from flushConstantPool.

2013-11-20T14:02:27+00:00

https://bugs.webkit.org/show_bug.cgi?id=124278

Patch by Julien Brianceau  on 2013-11-13
Reviewed by Michael Saboff.

Random crashes may occur with sh4 architecture, when a flushConstantPool occurs in
movlMemRegCompact. As in this case a branch opcode and the constant pool are put
before the movlMemRegCompact, the branch itself is patched when calling repatchCompact
instead of the mov instruction, which is really bad.

* assembler/SH4Assembler.h:
(JSC::SH4Assembler::repatchCompact): Handle this specific case and add an ASSERT.

Change-Id: I9c0e78cade4d20d0d83d683ffe6a499cee63bdbb
git-svn-id: http://svn.webkit.org/repository/webkit/trunk@159203 268f45cc-cd09-0410-ab3c-d52691b4dbfc
Reviewed-by: Allan Sandfeld Jensen

[sh4] Fix load32WithUnalignedHalfWords function in baseline JIT.

2013-11-20T14:02:19+00:00

https://bugs.webkit.org/show_bug.cgi?id=124233

Patch by Julien Brianceau  on 2013-11-12
Reviewed by Michael Saboff.

* assembler/MacroAssemblerSH4.h:
(JSC::MacroAssemblerSH4::load32WithUnalignedHalfWords): Do not claim scratch register too early.
Test already covered by fast/regex/pcre-test-1.

Change-Id: Ib783d70723754e80d961ade9463d7dc85ccb83f0
git-svn-id: http://svn.webkit.org/repository/webkit/trunk@159153 268f45cc-cd09-0410-ab3c-d52691b4dbfc
Reviewed-by: Allan Sandfeld Jensen

[sh4] Refactor jumps in baseline JIT to return label after the jump.

2013-11-20T14:02:14+00:00

https://bugs.webkit.org/show_bug.cgi?id=123734

Patch by Julien Brianceau  on 2013-11-04
Reviewed by Michael Saboff.

Current implementation of jumps in sh4 baseline JIT returns a label on the jump itself
and not after it. This is not correct and leads to issues like infinite loop the DFG
(https://bugs.webkit.org/show_bug.cgi?id=122597 for instance). This refactor fixes this
and also simplifies the link and relink procedures for sh4 jumps.

* assembler/MacroAssemblerSH4.h:
(JSC::MacroAssemblerSH4::branchDouble):
(JSC::MacroAssemblerSH4::branchTrue):
(JSC::MacroAssemblerSH4::branchFalse):
* assembler/SH4Assembler.h:
(JSC::SH4Assembler::jmp):
(JSC::SH4Assembler::extraInstrForBranch):
(JSC::SH4Assembler::jne):
(JSC::SH4Assembler::je):
(JSC::SH4Assembler::bra):
(JSC::SH4Assembler::linkJump):
(JSC::SH4Assembler::relinkJump):

Change-Id: Ia31b326a1f3c41e9c913e513563ff1e6b8f8e55e
git-svn-id: http://svn.webkit.org/repository/webkit/trunk@158580 268f45cc-cd09-0410-ab3c-d52691b4dbfc
Reviewed-by: Allan Sandfeld Jensen

Fix compilation on ARM

2013-11-18T12:24:45+00:00

Android fix broke ARM compilation on non-Android Linux platforms.
Now correcly define Elf32_auxv_t only on Android/Qt and otherwise
include asm/hwcap.h

Same fix was previously applied to QtScript, which had the exact
same android problem (QTBUG-30978).

Task-number: QTBUG-34917
Change-Id: I29d570731754c6291f8dc0c379e6fdc8ec72b897
Reviewed-by: Allan Sandfeld Jensen

Only use 64bit atomics when supported

2013-10-28T14:42:33+00:00

To support Windows XP we must avoid using the 64bit variants of the
atomicIncrement/atomicDecrement operations which are only supported from
Windows 2003 and up.

This patch rolls back the minimum Win32 API version to WinNT 5.1 (32bit WinXP),
and adds a new flag to indicate the support of 64bit atomics.

The new flag is now also used to support MIPS and other architectures
without 64bit intrinsics, instead of hacking in poor support.

This also extends the atomic operations to sparc64 which was previously
skipped because it did not work with __exchange_and_add.

Task-number: QTBUG-34271
Change-Id: I21b09df3cafb7f0987a2f44c89036ff34ed34aa0
Reviewed-by: Jocelyn Turcotte 
Reviewed-by: Kai Koehne

Fix build for boot2qt eAndroid

2013-10-21T21:56:18+00:00

Reverts the removal of Android support in WTF, and brings the support
up to date.

Merged change to MacroAssemblerARM.cpp from QtScript.

Replaces use of statvfs with statfs in one place.

Replaces shm-based shared memory with ashmem-based in WebKit2.

Change-Id: I440b1fbd94bb4148f7ba764d77de65230d13ed90
Reviewed-by: Simon Hausmann